1. Massive MIMO and
Channel Modeling for Millimeter Wave
Gustavo Fraidenraich
Engenharia Elétrica
Departamento de Comunicações
Unicamp
1
2. Achieving 10000x capacity
Source: IEEE Spectrum, July 2004, n. 7 2
10x
Performance
20x
Spectrum
50x
Base Stations = 10000x
Performance
Massive MIMO mmWave Densification
3. What is Massive MIMO?
T. L. Marzetta, “The case for MANY (greater than 16) antennas as the base station,” in Proc. ITA, San Diego, CA, USA, Jan. 2007.
Thomas L. Marzetta , "Noncooperative Cellular Wireless with Unlimited Numbers of Base Station Antennas ,” IEEE Trans. Commun. 2010.
BS
User 1
User 2
User K 3
M
M-1
1
2
4. 4
Antenna Array Gain
1 Element
1.0
0.5
0.0
-0.5
1.0
0.5
10
1.0
0.5
0.0
-0.5
10 Elements 20 Elements
-1.0 -0.5 0.0 0.5 1.0
20 Elements
-1.0 -0.5 0.0 0.5 1.0
-1.0
N=1
0.0
-0.5
-1.0
-1.0 -0.5 0.0 0.5 1.0
1.0
0.5
0.0
-0.5
-1.0
20
-1.0 -0.5 0.0 0.5 1.0
-1.0
5
2 Elements
Antenna Aperture λ / D
D
5. 5
What is Massive MIMO
Essentially multiuser MIMO with lots of base station antennas
Hundreds Tens of Users of BS antennas
A very large antenna array at each base station
A large number of users are served simultaneously
An excess of base station (BS) antennas
6. 6
Maximal Ratio Combining
Uplink
BS
User
M
*
1
2
*
h1
*
h2
hM
Σ
h2
h1
hM
7. 7
Maximal Ratio Transmission
Downlink
BS
User
M
1
2
h2
h1
hM
*
*
*
Knowledge of the Channel at the transmitter side.
Reciprocity!
h1
h2
hM
8. 8
Bit Error Probability
Maximal Ratio Combining
y = x + z
Pb = Q
2Eb
N0
⎛
⎝ ⎜
⎞
⎠ ⎟
y = [hhh!h]x + z
1 2 3M y = hx + z
h†y
MRC
M
Pb = 1
2
1− γ b
γ b + M
⎛
⎞
⎟
⎠ ⎜⎝ M −1+ k
M−1 Σ 1
k
⎛
⎝ ⎜
⎞
⎠ ⎟
k=0
2
+ 1
2
γ b
γ b + M
⎛
⎝ ⎜
⎞
⎠ ⎟
k
AWGN Channel
AWGN Channel
+Fading with
γ Diversity b = Eb
N0
9. 9
Maximal Ratio Combining
Bit Error Probability
0 5 10 15 20
1
0.1
0.01
0.001
10-4
10-5
10-6
M=1
M=2
M=8
M=50
Only Gaussian
Noise
17 dB
12. 12
System Model
h1
h2
hK
x1
x2
xK
Processing for user i
KΣ
y = xihi
i=1
+ z
*y
M
1
M
hi
* →1
hihi
1
M
* →0
hihj
13. 13
MRT Precoding
MASSIVE MIMO FOR NEXT GENERATION WIRELESS SYSTEMS
Erik G. Larsson, ISY, Linköping University, Sweden Ove Edfors, Lund University, Sweden Fredrik Tufvesson, Lund University, Sweden Thomas L. Marzetta,
Bell Labs, Alcatel-Lucent, USA
15. System Model
15
x
S3 Multipath
h n
15
Slow Fading +Shadowing
Fast Fading
16. 16
Signal-to-interference-plus-noise Ratio
SIR = β jkl
2
β jkl
2 +Gv
l≠j Σ
⎯M⎯→⎯∞→ β jkl
2
β 2
jkl
l≠j Σ
M
β 2
jkl
l≠j Σ
Gv
• Fading and noise vanish as M grows to infinity!
• SIR expression is independent of the transmitted powers.
• For an arbitrarily small transmitted energy- per-bit, the SIR can be
approached arbitrarily closely by employing a sufficient number of
antennas.
19. 19
Experimental Results for Massive MIMO
Lund University - Sweden
128 antennas freq. 1.2 ~ 6 GHz
10 users
National Instrument Plataform - USRP
1,2 meters
20. 20
Experimental Results for Massive MIMO
Lund University - Sweden
High speed data streaming for multiple users
10 mobile uses stream HD
video on uplink to basestation
Basestation streams 10 HD
videos on downlink to users.
21. 21
Experimental Results for Massive MIMO
Lund University
128 Antennas 128 Virtual Antenna Array
22. 22
γ = λmax −λmin
γ
4 Terminals, M=4,32, and 128 - H (4 x M)
23. 23
Experimental Results for Massive MIMO
LOS scenario with four
users co-located
NLOS scenario with four
users co-located
LOS scenario where
the four users are
well separated.
Angle of Arrival
24. 24
Experimental Results for Massive MIMO
Argos: Practical Many-Antenna Base Stations
Rice University, Bells Labs and Yale University
64 Antennas
WARP Plataform
freq. 2.4 GHz
Argos: Practical Many-Antenna Base Stations
Clayton Shepard, Hang Yu, Narendra Anand, Lin
Zhong1
Li Erran Li, Thomas Marzetta2,
Richard Yang3
26. 26
#
! !"
&
=
$ $%
h h
11 12
h h
21 22
H
11 h
22 h
21 h
12 h
MIMO Model
Mt Mr
C = min Mt ,Mr ( )log2 (1+ SNR)
if Mt ≫ Mr
C = Mr log2 (1+ SNR)
Capacity scales with the number of users
31. Channel Modeling for millimeter Wave
• Parameters
– Free Space Attenuation
– Path Loss Exponent
– AOA (Angle of Arrival) and AOD (Angle of
Departure)
– Penetration loss
32. 32
Free Space Attenuation
The equation often leads to an erroneous belief that free space attenuates an
electromagnetic wave according to its frequency.
The expression for FSPL actually encapsulates two effects:
Distance dependency Frequency dependency
of Antenna
Attenuation = PT
PR
= 4π d2 4π f 2
c2
1
G
Antenna Gain=1
33. 33
Free Space Attenuation
d=150 m
10 100 1000 104
60 GHz
d HmetersL
AttenuationHdBL
140
120
100
80
60
3 GHz
26 dB
d=3000m
A(dB) = 20log10
4π
c
df ⎛⎝ ⎜
⎞⎠ ⎟
= 20log10 (d)+ 20log10 ( f )−147.55
f - Hz
d - meters
Antenna Gain = 1
34. 34
λ −2
Free Space Attenuation
• For a fixed antenna area, the beamforming gain grows with ;
• The increase in path loss can be entirely compensated by applying beam
forming;
• In fact, the path loss can be more than compensated relative to today’s
cellular systems, with beamforming applied at both ends.
• We conclude that maintaining the same physical antenna size, mmW
propagation does not lead to any reduction in path loss relative to
current cellular frequencies.
35. Path Loss Exponent
L =10nlog10 (d)
180
135
90
45
0
n=6 - Indoor Environments
n=4 - Two Ray Model
n=2 - Free Space
n=1,5 Waveguide
1 10 100 1000
d (meters)
L (dB)
43. AOA - Angle of Arrival
1) As the frequency increases, decreases and the
therefore the resolvability of the antenna array increases.
2) As the frequency increases the angular spread decreases.
43
θ ~
λ
D
Source: David Tse book
44. 44
AOA - Angle of Arrival
28 GHz
6 main Lobes
George R. MacCartney Jr and Theodore Rappaport, "Millimeter Wave Propagation Measurements for Outdoor Urban
Mobile and Backhaul Communications in New York City,”IEEE ICC 2014.
45. George R. MacCartney Jr and Theodore Rappaport, "Millimeter Wave Propagation Measurements for Outdoor Urban
Mobile and Backhaul Communications in New York City,”IEEE ICC 2014.
45
AOA - Angle of Arrival
73 GHz
3 main Lobes
46. 46
AOA - Angle of Arrival
In order to overcome the loss in the degrees of freedom, we must
use 2D antennas.
47. 47
Delay Spread
The RMS delay spread is independent of frequency in the LOS scenario
Source: Dajana Cassioli, Luca Alfredo Annoni and Stefano Piersanti, “Characterization of Path Loss and Delay Spread of 60-GHz UWB Channels vs. Frequency, “ IEEE ICC 2013 - Wireless Communications
Symposium.
48. 48
Delay Spread
For NLOS, delay spread increases with the frequency and then
saturates.
49. 49
Set of measurements at 10 GHz
- Penetration loss
- AOA
- Knife edge diffraction
- Delay Spread
Prof. Matti Latva-Aho
PhD. Student Claudio F. Dias
56. Wall Penetration Loss Measurements
56
• Simple penetration loss
measurements with few
antenna locations
• Idea was to measure the
penetration by moving
antennas only fractions
of wavelength between
the measurements
57. 57
Conclusions
Benefits from the (many) excess antennas
Simplified multiuser processing (MRC and MRT)
Reduced transmit power
Thermal noise and fast fading vanish
mmW Communication
Narrow-beam communication is new to cellular
communications and poses difficulties.
Free space does not increase as frequency increases
(keeping the same effective antenna area).
Penetration loss is the new problem (on-off behavior of
the channel).
The loss of degrees of freedom, as frequency increases,
may be compensated using 2D antennas.
We need 3D channel modeling to better understand all the
physical phenomena.
58. 58
References
[1] - Mustafa Riza Akdeniz, Yuanpeng Liu, Mathew K. Samimi, Shu Sun, Student Member, IEEE, Sundeep Rangan,
Theodore S. Rappaport, and Elza Erkip, "Millimeter Wave Channel Modeling and Cellular Capacity Evaluation,”, IEEE
JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 32, NO. 6, JUNE 2014.
[2] - Millimeter Wave Cellular Ultra-Wideband Statistical Channel Model for NonLine of Sight Millimeter-Wave Urban
Channels Communications: Channel Models, Capacity Limits, Challenges and Opportunities
Prof. Ted Rappaport NYU WIRELESS, NYU Polytechnic School of Engineering, Joint work with Sundeep Rangan and Elza
Erkip.
[3] - A. F. Toledo, D. GJ Lewis, and A.M.D. Turkmani, "Radio Propagation into Buildings at 1.8 GHz”
[4] P. Nobles, and F. Halsall, "Delay Spread and Received Power Measurements within a Building at 2GHz, 5 GHz and 17
Ghz,”
[5] - Maria-Teresa Martinez-Ingles, Davy P. Gaillot, Juan Pascual-Garcia, Jose-Maria Molina-Garcia-Pardo, Martine Lienard,
and José-Víctor Rodríguez, “Deterministic and Experimental Indoor mmW Channel Modeling, “IEEE ANTENNAS AND
WIRELESS PROPAGATION LETTERS, VOL. 13, 2014 1047.
[6] -D. Cox, "Measurements of 800 MHz Radio Transmission
Into Buildings with Metallic Walls”, The Bell System Technical Journal 1983
[7] - A. F. Toledo, , Adel Turlmani, and David Parsons, "Estimating Coverage of Radio Transmission into and within
Buildings at 900, 1800, and 2300 MHz,” IEEE Personal Communications April 1998.
[8] - Hao Xu, Member, IEEE, Vikas Kukshya, Member, IEEE, and Theodore S. Rappaport, Fellow, IEEE , “Spatial and
Temporal Characteristics of 60-GHz Indoor Channels, “IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL.
20, NO. 3, APRIL 2002.
[9] - Mathew Samimi, Kevin Wang, Yaniv Azar, George N. Wong, Rimma Mayzus, Hang Zhao, Jocelyn K. Schulz, Shu Sun,
Felix Gutierrez, Jr., and Theodore S. Rappaport , 28 GHz Angle of Arrival and Angle of Departure Analysis for Outdoor
Cellular Communications using Steerable Beam Antennas in New York City, VTC 2013.
[10] - Theodore S. Rappaport, Yijun Qiao, Jonathan I. Tamir, James N. Murdock, Eshar Ben-Dor , “Cellular Broadband
Millimeter Wave Propagation and Angle of Arrival for Adaptive Beam Steering Systems (Invited Paper),”RWS 2012.
[11] - Dajana Cassioli, Luca Alfredo Annoni and Stefano Piersanti, “Characterization of Path Loss and Delay Spread of 60-
GHz UWB Channels vs. Frequency, “ IEEE ICC 2013 - Wireless Communications Symposium.